[Question]: Vector dimension error
Do you need to ask a question?
- [x] I have searched the existing question and discussions and this question is not already answered.
- [ ] I believe this is a legitimate question, not just a bug or feature request.
Your Question
Hi. I'm running LightRag using Docker Composer. The first time I ran it, I had the wrong embedding size of 3072. Then I changed it to 1024 in the .env file. After restarting, I load the document and get an error. Raw response content: b'{"status":{"error":"Wrong input: Vector dimension error: expected dim: 1024, got 3072"},"time":0.020959189}'
I deleted the working directory and the Docker volumes. I still get the same error. I don't understand where the information is saved. Has anyone encountered this?
Additional Context
No response
Recreate the Qdrant collection (or change the collection name) so its Size is 1024, or purge LightRAG’s working_dir if using NanoVectorDB, then reinsert the document.
This depends on the vector database you are using. For certain databases, such as PostgreSQL, manual deletion of the table or database instance is required.